Research

Racial Discrimination Against Crime Victims: Evidence from Investigations of Violent Crime (Job market paper) 

I develop and apply a new test for a consequential but overlooked form of racial discrimination: that against victims of crime, by police detectives. I test whether detectives display racial bias using a marginal outcome test derived from a model of police investigations. By performing the test on novel data on all the investigations of violent crime conducted by the Chicago Police Department over the last two decades, I find evidence of racial bias against Black victims: cases that are classified as solved in the same amount of time are more likely to be rejected by the prosecution for insufficient evidence if the victim is Black rather than White. This is true for both homicides (-9.2%) and non-fatal violent crime (-2.5%). These differences are consistent with the police being more tolerant of a case failing to enter the court process when the victim is Black. Heterogeneity analysis leveraging the experience and race of the primary detective indicates that the results are driven by racial animus and not by inaccurate beliefs. These results suggest that efforts to increase oversight and transparency in policing should also extend to police investigations and the broader differential treatment of crime victims.

The Effect of Police Management on Crime and Officers' Behavior: Evidence from Compstat 

With Maria Mercedes Ponce de Leon

Abstract: We revisit the importance of policing management practices as a determinant of crime and officers' behavior, a relevant issue that has been under-studied so far. To do so, we leverage the staggered introduction by over 100 large US police departments of Compstat, a management system based on regular, data-driven, performance reviews of middle managers. Using a dynamic difference-in-differences approach, we show that Compstat has helped bring down crime rates by around 10%, with both violent and property offenses significantly decreasing. The main identified mechanism is deterrence, achieved through improvements in crime mapping and incentivized by regular performance reviews. However, the reliance of departments adopting Compstat on quantifiable statistical indicators seems to incentivize officers to increase arrests of civilians for minor offenses. We provide suggestive evidence that these additional arrests do not play a significant role in the crime decline and therefore constitute a welfare loss. This is especially true for Black residents, as they experience a disproportionate increase in such arrests. Our results suggest that Compstat is an effective strategy to fight crime, but its adoption should be accompanied by regulation and/or oversight to avoid over-policing.  

The Causal Effect of Partisanship and Leadership on Police Lethal Force: Evidence from Sheriff Elections in the United States
With Fu Jin
[Draft available upon request]

Abstract: What explains stark differentials in civilian fatality rates across police departments in the United States? We leverage quasi-random variation of close sheriff elections in a Regression Discontinuity Design (RDD) to estimate the causal effect of partisanship and leadership on policing culture and usage of lethal force. We find that Republican led sheriff departments have on average 1.72 more casualties per elected term (four years) relative to their Democratic counterparts, with effects being amplified and more precise in rural counties and under circumstances where officers perceive a threat to their lives. We also find that Republican led sheriff departments utilize significantly fewer cameras on officers and squad cars during policing activities. Our collective evidence suggests that Republican sheriffs favor less video evidence in order to provide greater latitude for deputies to more readily default to lethal force in dangerous encounters.

Racial discrimination against crime victims: Evidence from 911 Calls

The Disparate Effects of Police Consolidation on Clearance Rates

The Effect of Closing Public Mental Health Clinics on Crime